Shedding Light or Shooting in the Dark – How to define Fake News?

A proposed legislation against the “manipulation of information” is currently under consideration by the French parliament. It was voted by the National Assembly and rejected by the Senate, where Emmanuel Macron’s political party does not hold a majority. A joint committee of both chambers has been summoned in the hope of finding an agreement. This bill aims – among other things – to create an emergency legal remedy. During the three months before a national election, a judge could be asked to stop within 48 hours the spread of “fake” or “false” news (fausses informations). This tool against misinformation raises a lot of questions, including procedural and technical ones.

I will only focus on a particular point: How do we define fake news?

Struggling for Words

The efforts made by the French parliament perfectly illustrate the difficulty of this task, and the rather unsatisfying outcomes of these efforts result in the need to come up with a better definition. Every attempt to regulate fake news must indeed start from a sufficiently precise definition. The issue of fake news is more specific than the general issue of truth in electoral campaigns. Regulating fake news is not the same as restricting every inaccurate political statement.

The original bill contained no definition of fake news. A parliamentary commission defined them later as the “allegation of a fact lacking any verifiable element making it credible” (Toute allégation ou imputation d’un fait dépourvue d’éléments vérifiables de nature à la rendre vraisemblable). This definition was dropped in plenary session and replaced by another one, defining fake news as “any allegation of a fact that is inaccurate or misleading” (Toute allégation ou imputation d’un fait inexacte ou trompeuse). Furthermore, the judge shall put a halt to the spreading of such fake news only if they are likely to “distort the fairness of the election” (de nature à altérer la sincérité du scrutin), and if their propagation on the internet was made “deliberately” and “in an artificial or automatized and massive way” (diffusées de manière délibérée, de manière artificielle ou automatisée et massive). The bill is not targeting the author of fake news but its massive propagation.

The current definition deals with the content of the speech, the consequences of its diffusion, and the intent of its propagator. The definition might change again in the future, but one can build on this proposal to search for a better definition of fake news which I will do after a scrutiny of the aforementioned elements.

From Misleading Opinions to Fake News – Content

Fact or Feeling?

Fake news are false. Every definition of fake news should clearly say so. The words used in the parliamentary definition (inaccurate or misleading) are therefore not the best choice. Furthermore, the conjunction “or” indicates that an accurate but misleading statement could qualify as “false information.”

The criterion of falsity implies that fake news are factual statements. Opinions can seem absurd or despicable, but they cannot be false. However, one should keep in mind that an expression dressed up as an opinion can imply factual elements: “If a speaker says, ‘In my opinion John Jones is a liar,’ he implies a knowledge of facts which lead to the conclusion that Jones told an untruth. […] Simply couching such statements in terms of opinion does not dispel these implications.”((SCOTUS, Milkovich v. Lorain Journal, 497 U.S. 1 (1990), p. 18.))

The criterion of falsity excludes satire, parody, hyperboles or similar rhetorical devices. “For better or worse, our society has long since passed the stage at which the use of the word ‘bastard’ would occasion an investigation into the target’s lineage or the cry ‘you pig’ would prompt a probe for a porcine pedigree.”((Levinsky’s v. Wal-Mart Stores, 127 F.3d 122 (1997), p. 128.)) Statements of facts must be reasonably understood as serious, referential affirmations. During the parliamentary debate, many examples were given to show the dangers and absurdity to regulate fake news. But most of them were not statements of facts. Here are a few of these examples.

During the campaign about the European Constitution, French politician François Bayrou stated that, if the “No” prevailed, it would rain during more than 40 days. With your statute, you would ban such an expression, said a representative during the recent debate. Another representative said at one point that “the road to hell is paved with good intentions.” Jean-Luc Mélenchon interjected: “Fake news!” But these expressions are metaphors, rhetorical hyperbole and the like. They are not statements of facts and can therefore not be described as false.

Crucial or Petty?

This aspect of fake news is hard to define precisely but it is nevertheless important. Some authors mention in this context the “materiality” of the false assertion.((See Clay Calvert ans Austin Winning, «Filtering Fake News Through a Lens of Supreme Court Observations and Adages», First Amendment Law Review, 2018, p. 158.)) The falsehood shall not deal with a minor element but with a fact of sufficient importance.

The word “inexacte” in the French definition is therefore inappropriate, since it can mean “false” but also “inaccurate.” However, this flaw might well be repaired with another element of the bill: the diffusion of the false information must be likely to “distort the fairness of the election” which can only be the case when the falsehood relates to a significant fact.

Fake or False?

This extremely important point was completely ignored by the French parliament. Every false statement of facts does not qualify as fake news, as manipulation of information, to quote the title of the French bill. The expression must masquerade as a real news report. In this sense it can be qualified as “fake” – and not simply “false” – news.

This is crucial. One should not be induced in error by the proliferation of the words “fake news” that are now used to criticize any purportedly false statement made by anyone. Regulating fake news does not amount to restricting any inaccurate or even plainly false statement during a campaign. Fake news imitates the style of a news report to spread falsehoods. The statement uttered by a political candidate or written in a campaign leaflet should not fall under a law against fake news.

For sure, giving a precise definition of the “appearance of real news” is not an easy task. But this does not necessarily mean that this criterion is very hard to check. News are like elephants and pornographic pictures: I cannot define it but I know it when I see it.

Obvious or Obscure?

This is the second serious gap in the French definition. Only unequivocally false statements can qualify as fake news, at least for the purpose of an emergency procedure like the one contemplated by the French legislator. The parliamentary proceedings show that the French parliament does in fact target such obvious false statements. But this element should appear in the statute.

Of course, this element brings a serious problem with it. The judge will decline to act against a statement that is not obviously false. Now, a statement is not true just because it is not obviously false. But the judgment might give rise to an impression of truth. Furthermore, proof of falseness might be hard to produce, especially when it is “negative”. I can easily prove that I paid my income tax. But how am I to prove that I do not own a bank account in Switzerland?

Candidates should think twice before bringing the matter before the judge. But one problem of the French bill is that it allows any person having an interest – which could mean any voter – to request a judicial decision.

I leave this important matter aside, because my interest here is not on the appropriateness to regulate fake news, but only on its definition. The second element of the definition deals with the consequence of the expression.

The Likelihood of Consequences

The bill targets the propagation of fake news only if they are likely to “distort the fairness of the election.” This point was strongly criticized by several representatives: how could the judge know before the election that an expression is likely to have such a consequence?

This criticism rests upon a misunderstanding. The judge will not have to show that the fake news will in fact impact the vote, nor of course will they have to show that the expression did impact a vote that has not yet occurred. The judge will have to assess the likelihood of the consequence. It is a quite usual process in the law governing freedom of expression, and not a very demanding one. The offence of incitement to violence is an example. Contrary to what prevails in US law, the French judge does not have to show a concrete and imminent risk of harm: a reasonable likelihood is usually sufficient.

One can bet that every fake news dealing with a candidate or a topic relevant to the campaign will be seen as satisfying this condition. As seen before, the real input of this element is to specify the content of the message, to circumscribe the topic of the false information.

Oops, I didn’t mean to – Intent

The bill targets the deliberate spreading of fake news likely to distort the fairness of the election. This element of intent therefore has a double aspect. The propagator shall know that the statement is false – or, in the current and hopefully soon to be dropped definition, “inaccurate or misleading” –, and they shall know that its diffusion is likely to have the dreaded consequence.

French law governing the freedom of expression shows that these conditions will not be difficult to meet. Intent is indeed examined through the standard of the reasonable man. If the statement is obviously false, the propagator should reasonably be aware of it. And if the judge considers that its diffusion is reasonably likely to affect the fairness of the election, so should the reasonable propagator.

Shedding Light on the Target

Starting from the French bill, the definition of fake news can be refined as follow. (Note that I remove the intentional element which plays a – relative minor – role in the legal regime of fake news, but is not necessary to the conceptual definition.)

The obviously false statement of fact having the appearance of real news and likely to distort the fairness of the election.

This definition can certainly be criticized, modified and improved. But one thing is certain: a sufficiently precise definition is a necessary preliminary condition to have a real debate on the specific issue of fake news. Without such a definition, the legislator will merely be shooting in the dark.

So far, the debate in the French parliament has been replete with pompous declarations on the freedom of expression, Orwellian references to the danger of a “ministry of truth” and so on. This, however, brushes aside the real questions and the specific problems raised by the issue of fake news.

It may be that a regulation of fake news is contrary to the French Constitution or to the European Convention on Human Rights. Furthermore, a restriction of fake news might be inopportune as it might raise more problems than it solves. There may even be nothing to fear and thus no need to legislate on fake news. As someone wrote, fake news might be a fake problem. While I did not deal with these issues here one cannot start to think about them without a precise definition of fake news. In the absence of a preliminary definition that sufficiently circumscribes its object, the French parliament runs the risk of having a fake debate.((These remarks were presented at the conference Misinformation in Referenda, organized in August 2018 at the University of Lausanne by Sandrine Baume, Véronique Boillet and Vincent Martenet.))